Cognitive theories of problem solving and suggestions made by cognitive psychologists regarding how to teach problem solving are reviewed. Theories and suggestions from creativity research are also considered. The results are summarized in terms of a description of how high levels of proficiency in problem solving are acquired and how problem solving skills might best be taught, keeping in mind a distinction between well- and ill-structured problems. The need for practice materials is discussed, and some desirable qualities of such materials are suggested. Finally, several unresolved issues regarding instructional methods are considered. (Author)